A method for real-time implementation of HOG feature extraction

Hai-bo Luo; Xin-rong Yu; Hong-mei Liu; Qing-hai Ding

doi:10.1117/12.901081

26 August 2011 A method for real-time implementation of HOG feature extraction

Hai-bo Luo, Xin-rong Yu, Hong-mei Liu, Qing-hai Ding

Proceedings Volume 8193, International Symposium on Photoelectronic Detection and Imaging 2011: Advances in Infrared Imaging and Applications; 819302 (2011) https://doi.org/10.1117/12.901081
Event: International Symposium on Photoelectronic Detection and Imaging 2011, 2011, Beijing, China

Abstract

Histogram of oriented gradient (HOG) is an efficient feature extraction scheme, and HOG descriptors are feature descriptors which is widely used in computer vision and image processing for the purpose of biometrics, target tracking, automatic target detection(ATD) and automatic target recognition(ATR) etc. However, computation of HOG feature extraction is unsuitable for hardware implementation since it includes complicated operations. In this paper, the optimal design method and theory frame for real-time HOG feature extraction based on FPGA were proposed. The main principle is as follows: firstly, the parallel gradient computing unit circuit based on parallel pipeline structure was designed. Secondly, the calculation of arctangent and square root operation was simplified. Finally, a histogram generator based on parallel pipeline structure was designed to calculate the histogram of each sub-region. Experimental results showed that the HOG extraction can be implemented in a pixel period by these computing units.

Citation Download Citation

Hai-bo Luo, Xin-rong Yu, Hong-mei Liu, and Qing-hai Ding "A method for real-time implementation of HOG feature extraction", Proc. SPIE 8193, International Symposium on Photoelectronic Detection and Imaging 2011: Advances in Infrared Imaging and Applications, 819302 (26 August 2011); https://doi.org/10.1117/12.901081

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available